Monday, March 9, 2026

1:50 PM - 3:00 PM

London & New York

---

Details

Session Leads: Roderick Perez, Arild, Theresia, Odd This 80‑minute session convenes operator, vendor, and academic perspectives to demonstrate how agentic and multimodal AI integrate geoscience and reservoir engineering across the value chain (G&G → reservoir → wells → production) with a clear business case (cycle time, recovery, cost, risk). It frames why integration has been hard—divergent problem frames, data/scale mismatches, modeling cultures, and organizational silos—and what is new: multimodal agents that read seismic, logs, reports, schedules, and SCADA; surrogate models that de‑bottleneck simulation; and standardized data products/knowledge graphs that make handovers machine‑executable. Short talks deliver concrete use cases on both sides: seismic interpretation, generative geomodels, and structural uncertainty mapping; AI‑guided history matching, surrogate‑accelerated forecasting, and closed‑loop reservoir management. A moderated panel distills guardrails and value realization—calibration and physics audits, uncertainty and economics, governance/MLOps and data lineage, human‑in‑the‑loop and accountability—and addresses pitfalls (domain shift, false confidence, process/incentive misfit). Attendees leave with actionable integration patterns (hybrid physics–ML, agentic plan–act–check orchestration, shared subsurface data products), a minimum quality bar, and a skills roadmap for near‑term adoption. Considerations: ● Map the end‑to‑end workflow (G&G interpretation → static/dynamic modeling → wells/production → economics) and show how artifacts (faults, horizons, wells, schedules) are exchanged with traceability. ● Present operator, vendor, and academic case studies demonstrating surrogate‑enabled scenario screening, hybrid physics‑ML calibration, and closed‑loop optimization in real projects. ● Identify common pitfalls and guardrails—validation and physics audits, governance/model cards, human‑in‑the‑loop roles—and outline the skills and team structures needed for near‑term adoption.

---

Moderator